After you set up a connection to your model, configure the inputs for the model. The
inputs are the fields in OpenPages® that contain
the data that you want the model to process.
About this task
Ask your data scientist to provide you with the inputs that are needed for the
model you are using.
You can use one OpenPages object type per model or you can select all
object types to use multiple object types with the model.
You can either specify object fields or
a view as the input. When you select a view, the view contains the OpenPages fields that you want the model to process.
Except for chart relationship fields, all object fields and relationship fields in the view are sent
as input to the model.
- Select the Input type:
- Select View definition to send the fields from a view to the
model.
If you select this input type, make sure that you configure a view with the View
AI insights button after you configure this model. Users click the button to access the
insights from the model. For more information, see Adding a button to view AI
insights.
- Select Object fields to send the fields from a specified object
type.
- Select the Object type that you want to use as input to the
model.
If you selected Object fields as the input type, you can
select Select all object types if you want to use the model with multiple
object types. However, only System fields can be used as input to the model.
- If you want the user to run the model, select Manual. If you want
the model to run automatically when all the required fields are set, select
Automatic.
- Map model input fields to OpenPages
using the following steps:
- Depending on the service type you chose in the model configuration, use one of the
following options to specify at least one model input field to map (such as
Description):
- Choose a value from the Model input fields drop-down list.
- 9.1.3 and later If the service type of the model is
Extension, you can toggle Advanced JSONata and type in a JSONata expression
instead.
If the model payload requires specific field names, then use those names as the model input
field names; otherwise, use any descriptive name.
- Use one of the following options to choose the OpenPages fields that contains the input data that
you want to send to the model (such as Description (System
Fields:Description)).
- Choose Select a field and then choose a field from the
OpenPages fields drop-down list.
In this option the data comes from a
single field and sent to the input field.
- Choose Enter expression, click the icon inside the field, and then type
in content and OpenPages variables.
In this
option the data can come from multiple fields based on the number of variables you include in the
prompt. The typed content and the data pulled in from different parts of the object forms the data
that's sent to the input field.
- If needed, create a row for each input that your model needs. Click Add
input to add a row.
- For deployed prompts, the prompt variables that you define, follow a schema. The name of each
prompt variable is the model input field you specified earlier in this step. After creating your
prompt variables, you must deploy them to IBM Watson® Machine Learning. The following schema is the JSON payload after you deploy the prompt with prompt
variables:
{
"parameters": {
"prompt_variables": {
"<input field name 1>": "<Value 1>",
"<input field name 2>": "<Value 2>",
etc...
}
}
}
- For AI service, the following schema is supported for a request:
{
"input_data": [
{
"fields": [
"<field 1>",
"<field 2>",
etc....
],
"values": [
[
"<Value 1>",
"<Value 2>",
etc...
]
]
}
]
}
The following is an example of a request in AI service:{
"input_data": [
{
"fields": [
"Description",
"Account Type",
"Classification",
"Annualized Value",
"Value as of Date"
],
"values": [
[
"Details of current assets for High Oaks Bank financial reporting. Resources owned by the bank are tracked which provides future economic value.",
"Balance Sheet",
"Assets",
1254.23,
"5/6/2025"
]
]
}
]
}
-
Click Next and continue with Map
outputs.